Prevalence of drug-resistant pulmonary tuberculosis in India: systematic review and meta-analysis

Accelerated tuberculosis (TB) control efforts have been threatened by the emergence of Mycobacterium tuberculosis strains that are resistant to potent first-line drugs (drug resistant tuberculosis or DR-TB) [1–3]. In 2015, the World Health Organization (WHO) estimated 480,000 incident multidrug resistant TB (MDR-TB; resistance of both isoniazid and rifampicin) cases globally. With an estimated 79,000 MDR-TB cases, India along with the Russian Federation and South Africa accounted for 45% of the total notified combined MDR-TB and rifampicin-resistant (RR-TB) cases in 2015 [4].

The management of DR-TB is critical and based on laboratory confirmation of TB and a clear understanding of drug resistance aided by drug susceptibility testing (DST) to ensure accurate diagnosis and early intervention of appropriate treatment [1, 3, 5]. Currently, the WHO recommended treatment strategy for complex MDR-TB comprises of a minimum of 5 drugs (including an injectable aminoglycoside) and a protracted treatment period of 18 to 24 months [1, 2]. However, only 50% of patients worldwide with MDR-TB achieve successful completion of treatment, partially owing to high death rates (250,000 [range, 16,000–340,000] estimated deaths from MDR-TB/RR-TB in 2015) and loss to follow-up [2, 4, 6]. In India, only 46% patients with MDR-TB have been reported to achieve treatment success in 2015 (vs 48% patients who achieved treatment success in 2014) with 20% each of death and lost to follow-up [7, 8]. Further, worsening outcome of extensively drug-resistant TB (XDR-TB; resistance to at least one fluoroquinolone and injectable aminoglycoside in addition to MDR-TB) has been reported in 9.5% patients with MDR-TB in 2015 [4].

Prevention and control of drug resistance is therefore strongly recommended by the WHO through implementation of routine surveillance systems driven by systematic DST [3, 9, 10]. Nationwide survey conducted in representative populations using standardized patient stratification and employing quality-assured rapid diagnostic methods are fundamental to a strengthened surveillance [9]. The Revised National Tuberculosis Control Programme (RNTCP) endorses the WHO recommended Directly Observed Treatment, Short course (DOTS) and systematic surveillance in India. This initiative was introduced in 1997 and achieved nationwide coverage in 2006 [8, 11]. Improvements in RNTCP surveillance approach have been noted in the recent years and India accounted for 27% of the global TB notifications in 2014 (12% from private sectors) [3, 12]. However, India remains one of the six countries with an enormous MDR-TB burden that failed to implement a nationwide drug-resistance surveillance (DRS) and relies largely on a sub-national evaluation approach [3, 8].

Currently, published studies have reported the prevalence of DR-TB from region-specific data obtained from city or state government health facilities or private set-ups. Epidemiological interpretations from these studies are challenged by large variations in research methodology, patient selection, diagnostic methods, unclear definitions of retreatment as well as data analysis and reporting. Further, till date, there has been no attempt to consolidate these studies to derive pooled prevalence estimates of DR-TB and stratify the prevalence based on geographical distribution. The present study was therefore designed to provide pooled estimates for DR-TB (MDR-TB, pre-XDR and XDR-TB) in India through systematic review and meta-analysis of published studies conducted across two decades (1995 to 2015).

This systematic review and meta-analysis attempted to demonstrate the geographical distribution of DR-, MDR- and XDR-TB and identify the high-risk regions and populations based on an analysis of published studies in India over the past two decades. To the best of our knowledge, the present study is the first to investigate the prevalence of DR-TB in India using systematic review of published studies. Pooled estimates for the countrywide prevalence of DR-TB and MDR-TB revealed a worsening trend between the two study decades. The estimates for MDR-TB subgroups from the present study were higher than the national estimates reported by the RNTCP for the year 2015 (15%, previously treated cases; 2.2%, newly diagnosed cases) and WHO estimates for India (16%, previously treated cases; 2.5%, newly diagnosed cases) [4, 7]. Estimates presented by global or national control programs are based on samples from government centers comprising of potentially susceptible populations or populations where the infection appearance or recurrence is monitored regularly and treated optimally. Therefore, estimates generated from an analysis of these samples may not be a true representation of the TB population in the real-world [16–18]. The present meta-analysis was based on results from published clinical studies conducted pan-India, reporting data for diverse patient populations at varied set-ups that include government tertiary care hospitals (not covered under RNTCP), outpatient clinics, private multispecialty hospitals and district level RNTCP centers. The data therefore, effectively entails regional influences and different epidemiological factors contributing to drug resistance and does not involve selective sampling of patients. However, it should be noted that the prevalence rates reported in the current analysis potentially reflect the status among suspected isolates referred for resistance testing and may not be reflective of prevalence rates of resistance in general, which may be lower.

Interrupted or irregular TB treatments are the strongest determinants for acquired mono-drug resistance and promote the risk of bacterial mutations that eventually culminate in relapses and MDR-TB [19, 20]. Regional analysis for estimates of drug resistance showed that the burden of DR- and MDR-TB in all regions (North, South, West, East and Central) increased over the 20-year period. West India had the lowest prevalence of DR-TB in decade 1 which increased considerably making it the region with the highest number of DR-TB cases in the 2006 to 2015 decade. The prevalence of MDR-TB in this region also increased between the two decades and the prevalence of primary MDR-TB in newly diagnosed smear-positive patients was higher in this region. The 12 studies from West of India included metropolitan cities such as Mumbai, Pune and major cities from Gujarat, highlighting the rapid emergence of DR- and MDR-TB in over-populated urban locales. Increased risk of infection transmission due to crowding, inadequacies in community TB control programs and most importantly, the high variability in the anti-TB treatment regimens prescribed by doctors, particularly in the private sector are some potential factors attributable to this upsurge [21, 22]. High rates of MDR-TB in Mumbai have previously been reported in individual studies involving RNTCP outpatients from municipal wards [23] and patients from a multispecialty private tertiary care hospital [21]. In contrast to the bigger cities in India, the studies in Central and East zones included population from rural and smaller towns. Among other factors, sparse population, access to free and supervised government aided medical centers and limited access to multiple doctors (leading to lesser variability in treatments) can be associated with the relatively lower prevalence of DR- and MDR-TB observed in this zone [21]. However, an overall underreporting of the DR- and MDR-TB burden due to difficult geographical terrain that limits accessibility to healthcare resources and poor socioeconomic status should not be overlooked [21, 24, 25].

Resistance to fluoroquinolones among pre-XDR-TB cases had the highest nationwide prevalence as compared with the rates for second-line aminoglycoside injectables. Easy access and indiscriminate use of fluoroquinolone antibiotics for other common non-TB infections are the most predictable risk factors for the development of resistance to these second-line drugs [26–31]. Findings from case studies suggest that short-term monotherapy with any fluoroquinolone can result in acquisition of resistance in MTB leading to serious implications that include poor MDR-TB treatment outcomes [32, 33]. Although, the estimates for XDR-TB over the 20-year period was low, of concern are the high rates of resistance to fluoroquinolones which have been regarded as one of the risk factors for the emergence of XDR-TB [28, 31, 34, 35]. India’s big share (63%) in the private TB market volume for second-line drugs is another major contributing factor for the high fluoroquinolone resistance observed [36]. Taking into account the minuscular share of the more preferred injectable second-line drugs (1% as opposed to 96% for fluoroquinolones [along with amoxicillin/clavunate]), fluoroquinolones are most likely to be used as monotherapy or even add-on to first-line anti-TB therapy instead of their recommended use as a second-line drug. Such irregularities in the usage of second-line drugs in private sector result in inadequate treatment for MDR-TB adversely impacting treatment outcomes and emergence of resistance [36, 37].

Mono-drug resistance to isoniazid and streptomycin were recorded at high levels and resistance to ethambutol alone had the least occurrence in India across both decades. Resistance to multiple first-line drugs underscores the importance of the implementation of the quadruple drug regimen for initial phase of tuberculosis treatment as advocated by DOTS [38]. The high levels of streptomycin resistance may be suggestive of its irrational use in non-DOTS treatment regimens at government and private set-ups [17, 39, 40]. Further, analysis of resistant strains have considered mono-drug resistance to isoniazid and streptomycin as factors that drive the development and amplification of additional resistance [41, 42].

Overall, these results emphasize on the importance of reinforcing DST in all patients previously exposed to anti-TB drugs to understand the drug resistance pattern and judiciously dispense standard or individualized chemotherapy for resistant cases. There is an impending need to curb the indiscriminate use of second-line drugs and advocate judicious use of newer drugs among physicians at various medical care set-ups to achieve better outcomes in patients with MDR-TB. The high prevalence of MDR-TB reported in the present study signifies the critical gaps in current treatment regimens and the need for fortification with better formulations comprising of newer drugs that have a distinct mode of action. In a country like India, where functioning of healthcare system heavily relies on the private sector, the adoption of newer drugs into government approved standardized regimens should be propagated unanimously and operational activities should be closely monitored for proper execution.

Some limitations of the present analysis should be considered. As the articles included for prevalence estimation did not encompass all states of India, these results may not truly represent the magnitude of DR-TB burden in India and should be interpreted with caution. In addition, the cumulative estimations of prevalence using a random-effect model may not completely invalidate the heterogeneity between studies. There was also a lack of adjustment for potential confounding factors such as socioeconomic status, age, gender etc. that could influence estimates derived from several studies. Further, it should be noted that an assessment of publication bias or selection bias was not performed.

Few noteworthy observations based on the review of published studies include the lack of standardized methods for DST adopted across India. The use of phenotypic and genotypic assays largely varied in public and private set-ups and was contingent on factors such as cost-effectiveness, availability of resources and sustaining infrastructure at various centers across India. This variability in turn introduces several incongruities such as, absence of standard definition of drug resistance and its different types and concerns pertaining to quality control, sensitivity, and reproducibility of results and validity of the laboratory techniques and could potentially affect the estimates from this meta-analysis [34, 43–46]. These observations emphasize the need to promote establishment and expansion of government endorsed laboratories with improved infrastructure that are capable of carrying out high quality, reliable and rapid turnaround DST.

Another grey area identified was the discordant recording of patient or clinical isolate data, which highlights the need for a standardized collection and reporting technique to aid better clinical correlations and decision making in India [47]. Some variables that contributed to these include differences in study durations and treatment strategies adopted across different regions and set-ups [46]. It is a challenge to understand the extent of nonadherence to medications or the quality of drugs taken by the patients since many were not on RNTCP recommended DOTS therapy [16]. The growing private healthcare sector in India is a major area of concern since these establishments involve the use and distribution of huge quantities of anti-TB drugs, with non-standardized treatment regimens that are not vigilantly supervised for adherence and completion [48]. These practices often lead to treatment interruptions and drug resistance is a consequence. In addition, timely notifications and efficient recording of patient details are regarded as early markers of community TB scenario and greatly support public healthcare programs. Inadequacies in these systems are therefore suggestive of looming danger [49]. In 2012, the Central TB Division (CTD) in collaboration with National Informatics Centre (NIC) initiated the implementation of a web-based application called ‘Nikshay’ [49]. This application primarily intends to create a robust database of all TB patients across India and enables access of this information to key policy makers, monitoring authorities and researchers who can positively impact treatment outcomes in TB-infected patients. The Government of India has mandated all private and government health establishments (outside the coverage of RNTCP) to ensure timely onward communication of patient details for the Nikshay repository [8].

There also exists a dire need for more regulated nationwide DRS based on standard epidemiological methods in India. Currently, sub-national DRS studies have been conducted in Gujarat, Maharashtra and South of India and the RNTCP is in the process of steering a nationwide initiative [8]. The RNTCP jointly with the National Tuberculosis Institute, Bangalore; U.S. Centers for Disease Control and Prevention (CDC) and WHO have constituted a nationwide survey comprising of representative populations of newly diagnosed and previously treated pulmonary TB cases. This initiative is expected to provide estimates that will be more generalizable to the entire nation and assist evaluations against global figures for improved understanding of the overall TB health situation in India.

The pooled estimates from this study highlight the growing prevalence of DR- and MDR-TB in India that poses a new challenge to its clinical management and public health strategies. Future research involving assessment of clinical drug usage and identification of independent risk factors would be of great significance. Results from such studies along with robust prevalence estimates from the DRS may potentially help strengthen control measures, guide appropriate interventional and follow-up strategies in vulnerable populations and assist overall clinical decision-making.